PhD Candidate · Data Science · Worcester Polytechnic Institute

Oluseun Olulana

I study fairness in machine-learning ranking systems, with a focus on how large language models handle structured decision data. My research develops empirical methods to evaluate fairness, ranking utility, and the reliability of model behavior when sensitive attributes are missing or uncertain.

Curriculum vitae Google Scholar GitHub Email

Worcester Polytechnic Institute (WPI) · Expected graduation: 2026

Portrait of Oluseun Olulana

Research interests

Large language models & agentic AI

Exploring LLMs and tool-using agents for structured decision tasks, including reasoning, planning, LoRA fine-tuning, and retrieval-augmented generation (RAG) for domain-specific, knowledge-grounded applications.

Fairness in ranking

Fairness-aware ranking and re-ranking across classical and learned methods; exposure, representation, and utility trade-offs; and extending fairness from binary to multi-group and intersectional settings.

Missing sensitive attributes

Understanding how unknown or inferred demographic information affects ranking fairness.

Reliable evaluation

Reproducible experiments, sensitivity analysis, and stability of fairness outcomes.

News

  1. “Can LLMs Rank Candidates with Missing Sensitive Attributes Fairly?” accepted to TMLR.Paper
  2. “Can Large Language Models Rank Tabular Data Fairly?” accepted to IEEE ICSC 2026.Paper

Selected publications

  1. Can Large Language Models Rank Tabular Data Fairly?

    An empirical study investigating whether LLMs can perform fairness-aware re-ranking while preserving ranking utility in real-world tabular decision settings.

    Paper Code

  2. Can LLMs Rank Candidates with Missing Sensitive Attributes Fairly?

    Studies fairness-aware reranking when sensitive attributes are unavailable, comparing external inference, implicit inference, and an integrated inference-and-reranking approach.

    Paper Code

  3. Hidden or Inferred: Fair Learning-To-Rank with Unknown Demographics

    Examines how demographic inference errors affect fairness when sensitive attributes are unavailable, and the risks of relying on inferred demographic signals in ranking systems.

    Paper Code

Technical Strengths

AI research + practical ML engineering

🤖

Machine Learning & AI

Building and evaluating ranking systems with focus on fairness metrics, experimental design, ablation studies, and reproducible ML pipelines.

Scikit-learn PyTorch XGBoost TensorFlow
💬

Large Language Models

Designing robust prompts, fine-tuning language models with LoRA, evaluating in-context learning, and measuring cross-model stability for research applications.

GPT-4 Claude Llama OpenAI API LoRA
🐍

Research Engineering

End-to-end ML experiment pipelines, data processing workflows, API development, and interactive analysis for iterative research.

Pandas NumPy FastAPI Jupyter

Agentic AI

Exploring goal-directed AI agents that combine reasoning, planning, and tool use to carry out multi-step tasks and support research workflows.

Tool use Planning Workflow automation

Leadership

Selected service and leadership roles.

  • Member, WPI Data Science Council (2024-present)
  • Vice President, WPI African Graduate Student Association (2024-2025)
  • President, WPI African Graduate Student Association (2025-2026)
  • Parliamentarian, NSBE WPI chapter (2025-2026, partial term; stepped down to manage PhD workload)

Contact

Open to research and applied ML roles

Note: This site highlights academic and research work.